Introduction

Very few studies have been performed on the mass balance of glaciers situated between the tropics. The only studies so far published are of Lewis Glacier in Kenya Reference HastenrathHastenrath, 1984, Reference Hastenrath1989) and of two glaciers in the Cordillera Bianca. Peru (Reference AmesAmes, 1985; Reference Kaser, Ames and ZamoraKaser and others, 1990). These studies were conducted by collecting data on a yearly basis and were not coupled with systematic runoff measurements.

However, the coincidence at low latitudes between the wet season, favourable to accumulation, and the warm period, when ablation is maximum, makes it difficult to acquire knowledge of glacier mass balance from simple annual data. Furthermore, the precipitation distribution is fairly variable from year to year and the duration of the rainy season is yet more unpredictable. The glacier may be expected to react significantly to this variability.

Thus, if we want to understand the processes that control glacier mass balance in these regions, it is necessary to conduct measurements on a monthly basis and follow up the survey over a period of several years. With such an approach, it is possible to attain a greater understanding of the main factors that control the rapid glacier retreat presently occurring in this part of the Andes (Reference Lliboutry, Morales Arnao and SchneiderLliboutry and others, 1977; Reference AmesAmes, 1985; Reference JordanJordan, 1991), as well as in other low-latitude mountains (Reference Hastenrath and KrussHastenrath and Kruss, 1992). Another question needing to be addressed is the consequence of an ENSO (El Niño Southern Oscillation) event for the glacier mass balance. According to the study at the Quelccaya Ice Cap, Peru Reference Thompson, Mosley-Thompson and Morales ArnaoThompson and others. 1981. ENSO events could affect mass balance by decreasing snow accumulation at high altitude in the Andes. Nevertheless, the marked reduction in precipitation during these episodes occurs during the warm season, so die effects on the mass balance could also be associated with a strong ablation (Reference Francou, Ortlieb and MacharéFrancou, 1992).

Our investigation has concentrated on two glaciers in the Cordillera Real. Bolivia: Zongo Glacier and Chacaltaya Glacier. This paper presents the data collected on Zongo Glacier during the first two hydrological years of the study, from September 1991 to August 1993. It represents a first attempt to present a monthly pattern of glacier mass balance in the tropics and to analyse the effect of an ENSO event.

Area Description and Data-Gathering Methods

The glacier is situated in the Huayna Potosí massif (6088 m), at 16° S. in the Amazon basin. This valley-type glacier is 3 km long and has a surface area of 2.1 km (Fig. 1). The upper reaches are exposed to the south whereas the lower section faces east. The maximum and minimum elevations are 6000 and 4890 m a.s.l., respectively.

Fig. 1. Huayna Potosí and Zongo Glacier with the survey system in 1993. 1, Principal peaks; 2, Limits of basin; 3, Stakes; 4, Pits and crevasses; 5, Water-level recorder; 6. Rain-gauges: 7. Thermograph; 8. Pyranometer.

The accumulation zone amounts to 60% of the total area under steady-state conditions (annual mass balance approaching zero), at an equilibrium-line altitude (ELA) of 5100 m. Under similar conditions, average precipitation at the ELA totals 1.0 m a⁻¹ and the average temperature is −2 °C. Thus, it can be deduced that this glacier is mainly temperate with a limited cold surface near the summit.

A 15 stake network was placed in the ablation zone in order to determine the mass balance and the surface velocity, augmented by a further three stakes and several pits in the accumulation zone from 5500 to 5800 m (Fig. 1). The 15 stakes in the ablation zone were surveyed every month and the surface density of the snow and ice was measured in order to obtain the amount of accumulation and ablation in water equivalent, The accumulation in the upper reaches of the glacier was recorded immediately before and after the wet season, in September and April. Annual net balance was evaluated between the beginning and end of the hydrological year, September August. For the ice falls not surveyed by the stake network (the 5050–4950 and 5500–5300 m altitude ranges), net balance was estimated by linear interpolation. Mass balance was also computed independently by means of a hydrological method (Reference Ribstein, Tiriau, Francou and SaraviaRibstein and others, in press), whereb) we have the possibility of verifying the mass-balance values obtained by the glaciological method. In order for this to be done, a water-level recorder was set at 4830 m (150 m below the glacier terminus), providing continuous runoff measurements. Five storage rain-gauges, surveyed monthly, were placed near the glacier at 5200 and 4900–1850 m. These rain-gauges, with a cross-section of 2000 cm², have proved to be suitable for measuring snow precipitation and have shown an increase of 20% over the precipitation levels indicated by the lone classical 314 cm² rain-gauge in place since 1970 on Plataforma Zongo (4770 m).

In order to obtain estimates of the parameters controlling glacier ablation, thermographs with radiation shields were placed at 5200, 4830 and 4770 m. More recently, in 1993, pyranometers capable of measuring-global radiation were sited at 5200 and 4830 m. with a radiometer placed at 5200 m so as to record the net all-wave radiation.

Every year since 1991, a topographic survey has been conducted in August on the 15 ablation stakes and on the terminus contour to give estimates of the surface velocity and of the terminus fluctuation. During the first year of the study, the measured glacier retreat was 10–20 in, whereas during the second year the glacier terminus was stationary.

The easy accessibility of the glacier all the year round makes it possible to claim high levels of accuracy for data collected during the 2 years. But the progressive adjustment of the survey network during this time does not allow us to use all the collected data in this paper.

Monthly Net Balance in The Ablation Zone, 1991–93

The climates of 1991–92 and 1992–93 present very different patterns. The first year was marked by mature-phase ENSO conditions (Reference KouskyKousky, 1993). Values οΓ the south-oscillation index (the difference between the standardised sea-level pressure anomalies at Tahiti and Darwin) were strongly negative, with a minimum −2.2 standardised value for the 5 month running mean. This is the lowest monthly value since the 1982–83 episode, which has been described as a very strong event. Such a situation in the Pacific is associated with low precipitation levels in the central Andes. This observation was first pointed out in the Quelccaya Ice Cap by Reference Thompson, Mosley-Thompson and Morales ArnaoThompson and others (1984) and analysed further by Reference Francou and PizzaroFrancou and Pizarra (1985) and Reference Tapley and WaylenTapley and Waylen (1990) using data collected from meteorological stations.

The second year was quite normal. Temperatures recorded over a period of 30 a at the Chacaltaya station (5230 m), 10km south of Zongo Glacier, and precipitation collected over a 20 a period at Plataforma Zongo (4770 m) enable us to calibrate each month during both observed years Fig. 2). Thus, a clear positive deviation of the maximum temperature could be observed from January to Maj during the firs) year, linked with a higher radiation input and a deficit of rainfall. During a normal hydrological year. 7 months have more than 50 mm of precipitation, whereas in 1991–92 only 4 exceeded this total.

Fig. 2. Standard deviation of (a) maximum and minimum 1991–93 monthly temperatures from the 1953–93 mean at Chacaltaya (5230 m): (b) monthly 1991–93 precipitation from the 1972–93 mean at Plataforma Zongo (4770 m).

As presented in Figure 3. the monthly evolution of glacier balance in the ablation zone is more complex than may be inferred From the simple wet season dry season alternation typical of the tropics. During the ENSO event of 1991–92 a markedly negative net balance occurred in October December and March May, just before and after a short period of accumulation (January February). During the average year 1992–93. negative mass balance was limited to die beginning of the hydrological year, with April and May showing limited ablation. The variation in the balance from 5200 m to the terminus for every month of the first year of the study is presented in Figure 4a. Immediately before and after the January–March period the major part of the glacier experienced a strong ablation regime, even right up to high elevations. On the other hand, during a normal year such as 1992–93 (Fig. 4b), the monthly balance switched rapidly to a positive value. Whilst awaiting data over a longer period of time, we can already stress that October–December and March–May are decisive in the determination of the annual mass balance. In the central Andes, these are the months when variability of climate is at a maximum. However, it is obvious that the cold, dry months of June–August do have a limited influence at year scale, since the mass balance during these months approaches zero.

Fig. 3. Monthly balance from September 1991 to August 1993 in the ablation zone, calculated using two groups of stakes. Stakes 1–12 are situated near the FLA (5200–5100 m) and stakes 13-15 are near the glacier terminus (4900 m). The balance was estimated from a monthly stake reading and a density measurement of the snow/ice surface.

Fig. 4. Monthly net balance (α_m in m of water) as a function of altitude (Z) during (a) 1991–92, and (b) 1992–93. Numbers 1–12 refer to the months. Solid line: September, October, November, April, May. Dashed line: June, July, August. Dotted line: December, January, February, March.

Annual Net Balance and Water Discharge

For each year, the net balance β _n, has been calculated by integrating (he mass balance over the total surface area S, according to Equation 1

where S _e and S _a are the areas computed for the accumulation and ablation zones, respectively.

Over the two years, the specific net balance (α _n) per year was:

The ELA was situated in 1991–92 at 5300 m and in 1992–93 at 5100–5150 m (Fig. 5). In the same figure, we note that: (1) the curves are more-or-less parallel from one year to the other, as already observed on many middle-latitude glaciers, first by Reference Meier and TangbornMeier and Tangborn (1965) and Reference LliboutryLliboutry (1974); (2) during the ENSO event of 1991–92 the amount of accumulation a) high elevation is very low and this confirms the observations of Reference Thompson, Mosley-Thompson and Morales ArnaoThompson and others 1984) in Quelccaya. In such a case, even a glacier which like Zongo Glacier extends over a great altitude range may be almost completely submitted to an ablation regime.

Fig. 5. Annual net balance (ß _n) in m of water) as a function of the altitude Ζ during 1991–92 and 1992–93. It should be noted that the ELA was at 5300 m in 1991–92 and at 5100 m in 1992–93.

As indicated in Figure 6. die ENSO event 1991–92 was characterised by high runoff depths from October to December, with die December maximum 100 mm (33%) higher than during the following year. Furthermore, a second runoff peak occurred in March 1992. with runoff then remaining high in April and May. This confirms thai a marked deficit in the precipitation late in the warm season. March–April 1992, allows a rapid melting of the snow cover accumulated in January February. The water disc harge aí the Outlet of the 3 km² basin was 5.38 × 10⁶ m³ in the first year and 3.24 × 10⁶ m³ in the second. Considering the average precipitation (P) collected in the rain-gauges, the total precipitation received by the glacier was 0.916 m in 1991–92 and 1.060 m in 1992–93. We may assume that the area outside the glacier, 0.9 km², has a runoff coefficient of 0.8. This coefficient is quite important for a free glacier catchment in this Andean region (Reference Bourges, Ribstein, Hoorelbeke, Dietze, Cortez, Ricaldi, Flores and AnayaBourges and others, 1992). Henee, the water discharge corresponding to this area was 0.66 × 10⁶ m³ in 1991–92 and 0.76 × 10⁶ m³ in 1992–93. Thus, the amount of melted snow and ice corresponding to a 2.1 km² surface area is:

The specific net balance (ß _n) and the precipitation (P) being known, the total ablation (A) is deduced from Equation 2:

Fig. 6. (a) Extraterrestrial radiation and mean of hours per day with temperature above 3 °C, and (b) rainfall and monthly runoff depth, on Zongo Glacier during the 2 years of the study.

Consequently, it should be noted that the glaciological and hydrological balances give similar results. The loss by sublimation seems to be very small, lower than 10% of the total ablation. Considering the data accuracy, the sublimation cannot lie determined by means of the hydrological balance. This is consistent with the low values of sublimation obtained by Reference HastenrathHastenrath (1978) at the summit of the Quelceaya Ice Cap.

Factors Controlling The Glacier Balance

Whilst awaiting the opportunity to model the glacier ablation from recently installed temperature and radiation records, we can make a preliminary attempt to ascertain which variables are important in estimating the specific net-balance evolution in the ablation area of the glacier (i.e. between 5200 and 4890 m).

A stepwise multiple linear regression was applied to a large number of possible independent variables. From amongst all of the independent variables, the best result obtained pinpointed the following three as being the most critical:

• (1) the amount of radiation from the Sun arriving at the outer edge of the atmosphere, also called extra-terrestrial radiation: H _o in Wm⁻². This is constant at year scale, and obviously has a great influence at high altitudes at this low latitude. The monthly radiation values (Table.1 ; Fig. 6a) were calculated using the equations of Reference Paltridge and PlattPaltridge and Plate (1976);

• (2) the average daily duration of air temperature greater than 3°C at Plataforma Zongo (4770 m): (_d T ₃ in h d⁻¹ Table.1; Fig. 6a). Given the lapse rate of 0.74 ° C 100 m⁻¹ calculated between the two thermographs at 4770 and 5200 m. this 3°C temperature is equivalent to a positive temperature over practically all of the ablation zone;

• (3) the precipitation collected at 4770 m: P in m (Table.1; Fig. 6b).

With the specific net balance estimated in the ablation zone of the glacier (ß _na )) in m of water (Table.1; Fig. 6b) the linear regression using the monthly data is:

where r² is the correlation coefficient and n is the number of data.

It may be noted that _d T ₃ is the best correlated variable (r = −0.72).

With the runoff depth Q in m as the dependent variable (Table.1; Fig. 6b). the best regression was obtained using the H_o and _d T ₃ variables:

This time, the variable with the best correlation coefficient is H_o : r = 0.80.

These results suggest the following three comments:

• (1) Temperature and radiation closely control the ablation in the lower part of the glacier.

• (2) The greatest amount of solar radiation, leading to the melting of the greatest quantity of snow and ice, occurs near the time of the summer solstice. Therefore, we can begin to understand why a deficit in precipitation, and hence less frequent cloud-cover during this period, leads to the most significant melt. According to the observations of Reference Ribstein, Tiriau, Francou and SaraviaRibstein and others in press. the highest discharge recorded at the limnimetric station occurred altera 10d dry period during the warm season (October–March. It is dear that dry periods during the warm season are more frequent during ENSO events. It should be noted that the ablation produced during the dry season (April–September) with low radiation is not very significant at year scale.

• (3) Precipitation appears to be poorly correlated with glacier ablation and runoff, since in general the precipitation is of a solid form. Accumulated snow and hail have a melt time varying according to the climatic conditions. It might be possible at a later stage to determine these conditions.

Conclusions

For the first time, the collection of monthly data on an inter-tropical glacier allows us to highlight the most important factors that control mass balance at low latitudes. These factors are the duration of the wet season within the warm period, and the temperature level during the sensitive period which precedes and follows January-February, the height of the rainy season.

Hence, it can be noted that mass-balance determination is more complex for tropical glaciers than for their cousins at middle latitudes, for which only the three summer months appear to be critical. Since the climate in this part of the Andes is so variable, a longer-term study will be necessary to better understand this subject.

The second conclusion reached concerns the effect of the ENSO events. In comparison with the “normal” 1992–93 hydrological year. 1991–92 (an ENSO event was characterised by:

• (1) a markedly negative net balance, with a water depletion equivalent to twice the amount of the precipitation accumulated;

• (2) an increase in the elevation of the ELA by 200 m;

• (3) a reduction of the AAR (accumulation-area ratio) from 86% to 58%;

• (4) a very low accumulation rate at high altitude.

The above effects are a consequence of a marked increase of ablation due to a rise in the maximum temperature during the summer period, associated with increased radiation input, and also linked to the deficit in accumulation arising from a foreshortened season of 2 instead of the normal 4 months.

Thus the conclusions reached by Reference Thompson, Mosley-Thompson and Morales ArnaoThompson and others (1984) from the study of the Quelccaya Ice Cap concerning a decrease in the accumulation during ENSO events are confirmed. Furthermore, we can add that the ENSO effect modifies not only the accumulation term in the mass-balance equation, but also the ablation term, particularly in the present case study. The observations made during the 1991–92 ENSO event are confirmed by our analysis of water-level records from the Zongo Glacier outlet over a 20a period (Reference Ribstein, Tiriau, Francou and SaraviaRibstein and others, in press). The records reveal that the highest levels of runoff over this period were obtained during the three previous ENSO events:1976–77, 1982–83 and 1986–87), with a particularly high peak during the strong 1982–83 event.

The continuation of the present study over a period of several years would clearly be beneficial in furthering our understanding as to why tropical glaciers are currently experiencing such rapid retreat. Of further use would be a long-term study of mass-balance data from various other Andean glaciers.

It is already widely acknowledged that global atmospheric warming is a cause of glacier retreat. However, our study suggests that another important factor is the occurrence of short periods of warm, dry weather during the rainy season, a combination that is frequently linked to ENSO events.